Application of Neural Networks to Explore Manufacturing Sales Prediction

Round 1

Reviewer 1 Report

The paper has been improved following the referees’ suggestions, and now, according to me, it is ready to be published in this journal.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Instead of using only 35 indicators, including the net entry rate of employees into manufacturing industries, trend indexes, 21 manufacturing industry sales volume indexes, and customs export values authors need to consider using generative adversarial networks to generate large volumes of data.  Authors need to plot Accuracy, Recall, Precision, ROC, AUC, F1-score. The confusion matrix needs to presented and explained in great detail.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The paper proposes a prediction model, the propagation neural prediction method, to improve prediction accuracy for plastic injection machines sales in Taiwan. Results identify 7 key external economic factors.

As a general comment, I think the paper makes an interesting contribution to the literature by suggesting this new application of ANN. At the same time, I think the paper needs some minor improvements before to be published in this journal.

Broad comments

In the Literature Review, concerning the issue of application of Artificial Neural Network, I would suggest the author to cite the following papers where different technique about the forecast are compared Rafał Weron,“Electricity price forecasting: A review of the state-of-the-art with a look into the future”, International Journal of Forecasting, Volume 30, Issue 4, Pages 1030–1081, 2014 Silvano Cincotti, Giulia Gallo, Linda Ponta, Marco Raberto, “Modelling and forecasting of electricity spot-prices: Computational intelligence vs classical econometrics”, AI Communications, Volume 27, Issue 3, Pages 301-314, 2014 In order to let the paper more readable, I would suggest the authors to add a table with the main values used in the empirical analysis. I would suggest the authors to add a section where the empirical data used are described. I would suggest the author to justify why they can use the ANN with a small dataset (120 entries). The authors should give some more details about the methodology used. For example, how they divided the data in training and validation set.

 

Specific comments

In figures 2 and 3 I would suggest the authors to use not only different colour in order to plot the curves, but also different line style or marker so that also in black and white the figures are readable.

Figure 2 is not readable. Please improve the figure.

Reviewer 2 Report

Comments

Citation not provided in several statements Dataset details/distribution not provided Features correlations missing Lasks NN architecture diagram Please explain training and test in details Robustness check Unnecessary details like back propagation and pearson coefficient